cQ Analysis: Progress Report

Research purpose

The objective of this analysis is to first identify watersheds that might be more resilient to changes in water discharge and then to explore which variables (climate and land use) are contributing to that behavior.

Site Selection

For this nitrogen concentration and water discharge (cQ) analysis, we are using USGS data. The watersheds included in this analysis were gathered as the sites that meet the following criteria:

• There’s an active USGS gauge that monitors both discharge (00060) and nitrate concentration (99133).
• The gauge was installed prior to 2016, to get at least 6 full years of data.
• The watershed is inside the Upper Mississippi or Ohio basins (UMOB).
• The watershed is at least 400 square kilometers in size. This will be important for later stages where we want to compile spatial data.

This filtering criteria gets us the following 18 watersheds (map at the bottom).

Data Collection

Now that we have specific USGS gauges, we can collect the concentration and discharge data at a daily time step. Once the raw data is retrieved, we can analyze it by following these steps:

• Convert both variables into log, as c-Q has been observed to be linear in a log-log space.
• Calculate the slope of the cQ through a 30 day moving window. This will allow us to identify chemostatic and chemodynamic behavior at a daily level.
  • Enrichment: positive mx. As Q increases, c also increases
  • Dilutment: negative mx. As Q increases, c decreases
  • Chemostatic: slope near zero. Changes in Q do not affect c
• To verify the last step, we will also calculate Pearson’s correlation coefficient at each window-value. This will allow us to discard not-significant values.

Data from 2016 through 2021 will be used for this, giving us around 2,000 days.

Here are some graphics from these calculations:

The position of the points above can help us label the behavior in each point. For example, points in the top-right corner, which have a positive slope and positive correlation coefficient, are showing enrichment behavior. The points in the orange area have a slope near zero, and show chemostatic behavior. The gray bands are not-signficant values.

Clustering

In order to do a clustering analysis we need a standardized variable. We will use the percentage of chemodynamic days per year. This might give us an insight into which sites are more resilient to changes in discharge.